RevAIse Data Model Documentation¶
An open standard for describing, sharing, and reproducing AI-assisted systematic literature reviews
Welcome to the official documentation for the RevAIse Data Model.
Overview¶
RevAIse provides a comprehensive, standardized way to document every aspect of AI-assisted systematic reviews, ensuring:
- Transparency - Full documentation of AI usage, parameters, and decisions
- Reproducibility - Complete capture of methods, data, and computational environment
- Interoperability - Standard format for sharing and combining review data
- Traceability - Detailed provenance for all review stages and decisions
Quick Start¶
Access the Schema¶
The RevAIse schema is available in multiple formats:
- LinkML YAML - The source schema definition
- JSON Schema - For validation in applications
- JSON-LD Context - For linked data applications
Current Version¶
Version Information
You are viewing documentation for: 0.4.0
Release 0.4.0
ReadTheDocs automatically maintains documentation for all tagged releases. Use the version selector at the bottom of the page to switch between versions.
Key Components¶
Review Core Objects¶
These are the fundamental objects that characterize a systematic review:
- Review - The root container for systematic reviews
- Author - Review authors and contributors
- Protocol - Review protocol and registration details
- Literature Record - Individual literature items
Shared Infrastructure Objects¶
These objects are imported in review_core.yaml for sharing across stages:
- Registration Template - Templates for review registration
- Stage Execution - Base class for all review stages
- Stage Output - Outputs from stage executions
- Software Environment - Computational environment specifications
- External Tool - External tools and software used
- Enumerations - Controlled vocabularies and value sets
Review Stages¶
- Registration - Protocol registration and pre-registration
- Search - Literature search execution and documentation
- Screening - Title/abstract and full-text screening
- Extraction - Data extraction from included studies
- Synthesis - Data synthesis and meta-analysis
Features¶
Stage-Based Organization¶
Reviews are organized into discrete stages (registration, search, screening, extraction, etc.), each with: - Execution metadata (timing, actors, tools) - Input/output specifications - AI usage documentation - Quality control measures
AI Documentation¶
Comprehensive capture of AI assistance including: - Model specifications and versions - Prompts and parameters - Human oversight and modifications - Performance metrics
Provenance Tracking¶
Complete traceability with: - Temporal information for all activities - Actor attribution (human and AI) - Tool and environment specifications - Decision rationale
Quality Assurance¶
Built-in support for: - Review artifacts and checklists - Inter-rater agreement metrics - Conflict resolution documentation - Amendment tracking
Schema Formats¶
| Format | Description | Use Case |
|---|---|---|
| LinkML YAML | Source schema definition | Schema development and extension |
| JSON Schema | JSON validation schema | Application validation |
| JSON-LD Context | Linked data context | RDF and semantic web applications |
Version Support¶
This documentation system maintains all versions:
- Latest - The most recent stable release
- Dev - Current development version from main branch
- Tagged Releases - All historical versions (e.g., v0.1.0, v0.2.0)
Use the version selector to access documentation for any version.
Getting Started¶
- Explore the Schema - Start with the main schema documentation
- Review Examples - Check the schema for example instances
- Validate Your Data - Use the JSON Schema for validation
- Contribute - Visit our GitHub repository
Links¶
License¶
The RevAIse Data Model is released under the CC0 1.0 Universal license, making it freely available for any use.
Citation¶
If you use RevAIse in your work, please cite: